Cinematography Mailing List - CML
    advanced

Defining 2K & 4K

In the CML-Pro forum I have seen people referring to the terms "2K" and "4K" in a post production context. I have a couple of questions regarding this:

1 - What is 2K and 4K referring to and for what use? (Is it digital scans of film negatives for post work on computers? If so, are they referring to some kind of resolution? Can the "K" be referring to kilobytes? I would think one would need a much higher resolution for post work meant to go back to film or even TV for that matter).

2 - What books would you recommend for a detailed explanation of these terms and related technologies? I looked at the recommended book list on the CML website and found *Film Technology in Post Production, Second Edition *by Dominic Case... would this book explain it all?

Inquisitively,

Piotr Jagninski
Gaffer / New York City


>the CML website and found *Film Technology in Post Production, >Second Edition *by Dominic Case... would this book explain it all?

Piotr,

That is an excellent book! It will answer your question and then some.

Very simply, 4K and 2K refer to the amount of lines of resolution (K meaning thousand but not bytes). I'm probably not the best person on this list to answer your question, and even if it does get answered to your satisfaction I would more than recommend that book.

Nathan Milford
Future Gaffer
Brooklyn, NY


Nathan Milford wrote:

>I would more than recommend that book.

Me too.

Jeff "Hello, Dominic" Kreines


>1 - What is 2K and 4K

I'll take this one...

2K and 4K refer to the number of pixels across the width of the frame. So a full 35mm frame at 4K would be 4096 x 3112, while 2K is 2048 x 1556. Multiply that out by the bit depth (number of bits per colour per pixel) and you come up with nearly 50Mbytes per frame for 4K, and about 12Mbytes for 2K.

It's generally taken that 4K is the resolution needed to fully capture all the detail down to the level of individual grains, on 35mm film. 2k has been used more often though, as it's half the resolution for 1/4 the data and storage space. Either way, you're not wrong about 4Kbytes being inadequate for film or TV work. A totally uncompressed frame at standard TV res is worth about 1 Mbyte.

>*Film Technology in Post Production, Second Edition *by Dominic >Case... would this book explain it all?

Modesty prevents me from suggesting that your research has led you to the right book.

You could also have a look at www.atlab.com.au

Dominic Case
Atlab Australia


Dominic Case writes :

>It's generally taken that 4K is the resolution needed to fully capture all >the detail down to the level of individual grains, on 35mm film.

I think that with the latest stocks you may be looking at 6K.

I'll be publishing the test results in a couple of weeks.

Cheers

Geoff Boyle FBKS
Director of Photography
EU Based
www.cinematography.net


Dominic Case wrote :

>So a full 35mm frame at 4K would be 4096 x 3112, while 2K is 2048 x >1556.

4K and 2K refer roughly to the width of the scanned frame in thousands of pixels-- got it. Now by "full 35mm frame" I understand that to mean "full gate," as in film camera gates exposing the entire available negative (roughly sprocket hole to sprocket hole). Based on your pixel dimensions, am I correct in assuming then that full gate has an aspect ration of 1 : 1.32 ? And is that actually a meaningful number, ie. during production and/or post production does anyone really need to know the aspect ration of full gate?

How did we come up with the pixel width numbers 4096/2048 in the first place? Are these pixel widths now a standard to which all film digitizing (telecine?) machine manufacturers adhere?

>Multiply that out by the bit depth (number of bits per colour per pixel) >and you come up with nearly 50Mbytes per frame for 4K, and about >12Mbytes for 2K.

If I understood properly, to calculate the file size, in Megabytes, of a digitized full 35mm frame at "4K" one does the following: multiply pixel width(4,096) by pixel height(3,112) by the color bit depth. Correct? In the 4K/50Mbyte example you gave above, if I am not mistaken in my math, the color bit depth would have to be 4 (16 colors per pixel). If that is the case, is such a low color bit depth really good enough? Hell, my computer monitor at home handles 32 bit color. I would have thought the film industry would be a little more cutting edge, especially if the digitized frames were to be eventually returned to film for projection in big screen theatres. Am I missing something or wrong or is the technology simply not there yet?

>You could also have a look at www.atlab.com.au

I did go have a look...in case you are interested my web browser, Mozilla 1.5 (ostensibly 100% compatible with current internet standards), returns HTTP 404 Not Found errors all over the place when clicking most of your links, including the following :

* Sound Mixing
* News
* Film Technology in Post Production
* Digital Intermediates
* Members' information
* Welcome Announcement

Your site does seem to work with Internet Explorer however.

Mr. Case, if the answers to the questions I have asked above might not serve others in this forum and can be found in your book, please let me know.

Many thanks to all the people who responded to my previous query about 2K and 4K.

Cheers,

Piotr Jagninski
Gaffer / NYC


Piotr Jagninski writes :

>am I correct in assuming then that full gate has an aspect ration of 1 : >1.32?

Strictly 1.33:1, but a more direct way to obtain this information would be by reference to any of a range of reference books on cinematography (have a look at the list on the website www.cinematography.net/books.htm ), to get the actual standards.

>during production and/or post production does anyone really need to >know the aspect ration of full gate

They certainly need to know the aspect ratio (not ration, but I'm sure that's just a typo;-)) that the film is to be shown in. But if it's (say) a widescreen film in 1.85:1, then the area that will be scanned(e.g. at 4K) from the negative is less than 4096 since the soundtrack area is not used and less than 3112 because of the aspect ratio.

>How did we come up with the pixel width numbers 4096 (etc)

2x2x2x2x2x2x2x2x2x2x2x2x2

>all film digitizing (telecine?) machine manufacturers

The term telecine usually refers to machines that scan film and convert it to a video signal (analogue or digital). The highest resolution obtainable is HD which is 1920x1080 (although the colour information is only collected at 960 pixels wide).

Digitising to data, the machine is usually called a scanner. The "Spirit" operates in either mode and when in nominal 2K mode it's called a "datacine".

Different machines tend to have different CCD chips in them of different widths, so there isn't really a universal standard for scanning. However, data is stored in standard formats based on the 2K or 4K approach.

>in my math, the color bit depth would have to be 4 Am I missing >something or wrong or is the technology simply not there yet?

Rest assured that if your home computer can handle 32 bits then the professional motion picture industry has the technology too . . . What you are missing in your math(s) is the difference between bits and bytes.

At 10 bits _per colour_ (that is 30 bits, or about 100 million colours (not 16!), you'll find that the calculations work out right. Allow 8 bits per byte.

>my web browser, Mozilla 1.5 (ostensibly 100% compatible with current >internet standards), returns HTTP 404 Not Found errors

That's disturbing. We did test the site on a range of browsers, and occasionally bumped into some problems (though I don't recall any difficulties with Mozilla). however, I've taken the ignorant but pragmatic approach that the vast majority of people are using (or at least have access to) IE, and I can't cope (or afford to have someone cope) with all the variants. Glad you were able to get there with Mr Gates' help. (resist the urge to flame me, all you rebels out there).

Most of the above is indeed in my book. Go on - treat yourself, go out and look for it.

Dominic Case
Atlab Australia


Piotr Jagninski wrote :

>ration of 1 : 1.32 ? And is that actually a meaningful number, ie. during >production and/or post production does anyone really need to know the >aspect ration of full gate?

Meaningful - not really, but obviously useful for pan/scan, tracking and stabilisation in post amongst other things

>How did we come up with the pixel width numbers 4096/2048 in the first >place? Are these pixel widths now a standard to which all film digitizing >(telecine?) machine manufacturers adhere?

That would be Kodak's system 'fault' probably. The frame size for a Lightning 1 Scanner (Kodak's first motion picture one out in the real world) defines Full Aperture Full Resolution to be 18.672mm by 24.576mm give or take 0.012mm, which sampled at 4K (4096x3112) gives you a sensor capture size for the pixel of .006 mm - of course with magnification your actual sensor may be larger in area than this.

And yes these are generally accepted as the reference about which to build a system, e.g. Northlight also uses these as well as ARRI's laser recorder.

>If I understood properly, to calculate the file size, in Megabytes, of a >digitized full 35mm frame at "4K" one does the following: multiply pixel >width(4,096) by pixel height(3,112) by the color bit depth.

Almost, you forgot to multiply by the number of channels and add a bit for good look : 0 In the case of the Cineon/DPX files the bits you need to add take the 3 lots of 10 bits and pad them with an additional 2 bits to make 4 bytes per pixel. You then need to add headers and so forth so you know all the other picture information you might want which accounts for another 32K-512 for Cineon by default, 8KBytes for DPX.

>film for projection in big screen theatres. Am I missing something or >wrong or is the technology simply not there yet?

Something must have gotten messed up with your math - perhaps bits vs bytes as your out by a factor of 8

Roughly speaking scanned film is more like saying 40 bits per pixel in graphics card terms (the 32 bits per pixel you get on a PC is 8 bits per channel RGBA (A is for Alpha generally not part of your picture), but this is a bit of a 'whiter than white' kind of statement.

Comparing like with like, most graphics cards are 24 bit colour vs scanned films 30, so you need not worry about the result on screen - too much.

Something to take note of is that these are from systems built with some eye on practical usage. The resolution of Film is greater than 4K, the dynamic range is generally greater than 10 bits allows - but these specifications are at least 12 years old and the technology that was available at the time dictated some limits - even today working with 4K is a pain, and 2K is the normal size used.

Kevin Wheatley
Senior Do-er of Technical Things
Cinesite (Europe) Ltd


Dominic Case wrote :

>Most of the above is indeed in my book. Go on - treat yourself, go out >and look for it.

Mr. Case,

Your book just arrived today from (insert large book seller chain name here)... tonight I plan to make some tea, settle in and learn a few things!

Thank you for taking the time to answer my previous queries.

Cheers,

Piotr Jagninski
Gaffer / New York City


 


Copyright © CML. All rights reserved.